helpers/lxml: improve lxml_dump_node()

Closes #47

Author: stanislaw

reqif/helpers/lxml.py

@@ -7,7 +7,23 @@
 
 
 def lxml_dump_node(node):
-    return etree.tostring(node, method="xml").decode("utf8")
+    output = ""
+    node_no_ns_tag = etree.QName(node).localname
+    output += f"<{node_no_ns_tag}"
+    for attribute, attribute_value in node.attrib.items():
+        output += f' {attribute}="{lxml_escape_for_html(attribute_value)}"'
+    if node.text is not None or len(node.getchildren()) > 0:
+        output += ">"
+        if len(node.nsmap) > 0:
+            output += lxml_stringify_namespaced_children(node)
+        else:
+            output += lxml_stringify_children(node)
+        output += f"</{node_no_ns_tag}>"
+    else:
+        output += "/>"
+    if node.tail is not None:
+        output += lxml_escape_for_html(node.tail)
+    return output
 
 
 # This code is taken from Python 3.7. The addition is escaping of the tab
@@ -45,19 +61,24 @@ def lxml_escape_title(string: str) -> str:
 # FIXME: Would be great to find a better solution for this.
 def lxml_stringify_namespaced_children(node, namespace_tag=None) -> str:
     if namespace_tag is None:
-        assert (
-            len(node.nsmap) > 0
-        ), f"This method must be called on a namespaced tag:\n{lxml_dump_node(node)}"  # noqa: E501
+        assert len(node.nsmap) > 0, (
+            f"This method must be called on a namespaced tag. "
+            f"Tag: {node}, line: {node.sourceline}."
+        )
         nskey = next(iter(node.nsmap.keys()))
     else:
         nskey = namespace_tag
 
     def _lxml_stringify_reqif_ns_node(node):
         assert node is not None
-
         output = ""
         node_no_ns_tag = etree.QName(node).localname
-        output += f"<{nskey}:{node_no_ns_tag}"
+        tag = (
+            f"{nskey}:{node_no_ns_tag}"
+            if node.tag[0] == "{" or namespace_tag is not None
+            else node.tag
+        )
+        output += f"<{tag}"
         for attribute, attribute_value in node.attrib.items():
             output += f' {attribute}="{lxml_escape_for_html(attribute_value)}"'
         if node.text is not None or len(node.getchildren()) > 0:
@@ -66,7 +87,7 @@ def _lxml_stringify_reqif_ns_node(node):
                 output += lxml_escape_for_html(node.text)
             for child in node.getchildren():
                 output += _lxml_stringify_reqif_ns_node(child)
-            output += f"</{nskey}:{node_no_ns_tag}>"
+            output += f"</{tag}>"
         else:
             output += "/>"
 

tests/unit/reqif/helpers/lxml/test_lxml.py

@@ -1,24 +1,11 @@
 from lxml import etree
 
 from reqif.helpers.lxml import (
-    lxml_dump_node,
     lxml_is_self_closed_tag,
     lxml_stringify_namespaced_children,
 )
 
 
-def test_01_dump_xml() -> None:
-    spec_type_string = """\
-<PARENT>
-  <CHILD>text</CHILD>
-</PARENT>\
-"""
-    spec_type_xml = etree.fromstring(spec_type_string)
-    dump = lxml_dump_node(spec_type_xml)
-
-    assert dump == spec_type_string
-
-
 def test__02_stringify_namespaced_children__01_basic() -> None:
     expected_string = "\n<reqif-xhtml:div>--</reqif-xhtml:div>\n"
 

tests/unit/reqif/helpers/lxml/test_lxml_dump_node.py

@@ -0,0 +1,41 @@
+from lxml import etree
+
+from reqif.helpers.lxml import (
+    lxml_dump_node,
+)
+
+
+def test__01__basic() -> None:
+    spec_type_string = """\
+<PARENT>
+  <CHILD>text</CHILD>
+</PARENT>\
+"""
+    spec_type_xml = etree.fromstring(spec_type_string)
+    dump = lxml_dump_node(spec_type_xml)
+
+    assert dump == spec_type_string
+
+
+def test__02__xhtml() -> None:
+    expected_string = """\
+<THE-VALUE>
+    AA<reqif-xhtml:div>
+      11<reqif-xhtml:span attr="FOO">--<div>123</div>--</reqif-xhtml:span>22
+    </reqif-xhtml:div>BB
+  </THE-VALUE>
+"""
+
+    spec_type_string = """\
+<THE-ROOT xmlns:reqif-xhtml="http://www.w3.org/1999/xhtml">\
+  <THE-VALUE>
+    AA<reqif-xhtml:div>
+      11<reqif-xhtml:span attr="FOO">--<div>123</div>--</reqif-xhtml:span>22
+    </reqif-xhtml:div>BB
+  </THE-VALUE>
+</THE-ROOT>
+"""
+    spec_type_xml = etree.fromstring(spec_type_string)
+    string = lxml_dump_node(spec_type_xml.find("THE-VALUE"))
+
+    assert string == expected_string